Data processing: measuring – calibrating – or testing – Testing system – Of circuit
Reexamination Certificate
2000-12-07
2003-04-22
Nghiem, Michael (Department: 2863)
Data processing: measuring, calibrating, or testing
Testing system
Of circuit
Reexamination Certificate
active
06553329
ABSTRACT:
FIELD OF INVENTION
This invention relates to integrated circuit (IC) chip test software systems and more particularly to refining logic bitmap using pruned diagnostic list.
BACKGROUND OF INVENTION
Digital semiconductor integrated circuits are manufactured on dies of a silicon wafer. There are both memory chips and logic chips. A microprocessor chip is an example of a logic chip. A field known as Design for Test (DFT) involves designing into the chip “scan chains” that are used to write and read portions of the chip that might be otherwise inaccessible. Layout tools are used to lay out a chip design onto the silicon wafer. The chip layout may be represented in the form of a netlist which is a bit of low-level design cells and the interconnections between them. Once the design is completed, the bits of a file representing the chip are written to a tape or disk. A mask maker makes photomasks to manufacture the chip using the files.
The chips are production monitored and tested. Production monitoring is performed using “in-line” inspection equipment and production testing is performed using “end-of-line” test equipment. In-line inspection equipment inspects the entire semiconductor wafer, each of which may have hundreds of chips. The end-of-line test equipment performs what is called “binsort function test” on the semiconductor wafers wherein the pads of the chip are contacted by test equipment and the chips are operated. At the conclusion of the test, parts are “binned” or placed in different categories within an electronic record according to test result. The categories relate to the quality of the chip (bad, good, high quality, etc.).
Failure analysis attempts to identify the cause of failure of chips of a particular chip design after the failures have been detected during production or prototype testing. Failure analysis usually requires more detailed failure information than just a bin code. Detailed failure information is typically obtained by re-testing a number of packaged parts.
Memory chips in reality lend themselves to failure analysis because of their structure as regular arrays of memory cells. A memory chip may be tested by a series of read and write questions to the memory chip. The errors in read/write testing pin-point likely physical defects at readily identified location. Alternatively, the memory chip can have built in self test (BIST) capabilities. The functional test results can be “bitmapped” to failure locations on the memory chip. In memory bitmapping, electrical failures are realized within relatively small physical “trace” on the die.
FIG. 1
illustrates the process flow with conventional bitmap testing wherein the wafer undergoes in-line inspection (right column) and end-of-line testing (left-hand column). This in-line testing may be done using optical test equipment like KLA21xx from KLA Tencor to produce defect files containing X, Y location defect information. End-of-line testing is performed by Automatic Test Equipment (ATE), which identifies failed memory locations. This is processed to identify X, Y defect locations. The X, Y defect information may be output to a Yield Manager tool. Because these are in the same format within the Yield Manager tool, a combined defect overlay is obtained to determine “killer defects” or defects that render a part non-functional to be identified. The locations of a defect and a failure coincide. As a result of this identification, trouble-shooting would focus on the production process rather than the design of the part.
In recent years, the market share of logic products has increased with many new “logic only” semiconductor fabs without the benefit of a “memory production line to monitor the production process and take advantage of the yield enhancement technique developed within the industry.” There has been no way to “bitmap” area of logic within a chip. Furthermore, logic chip function test results do not provide a starting point for the physical coordinates of failures within a failed die.
The most advanced logic chip designs contain scan testing. Scan testing breaks the logic real estate of a chip into many discrete chains of logic which can be tested individually for basic functionality. Scan testing enables a list of failing signals to be identified. However, even after a list of failing signals is produced for a given dies, it is still not possible to find the physical location of the failure because each failing signal may contain hundreds of transistors within its “cone of logic,” and there are usually multiple failing signals. Traditional logic chip yield enhancement techniques therefore rely heavily on correlation of binsort functional test results to anticipate and correct semiconductor process issues. This approach suffers from several drawbacks, including: the inability to relate a particular bin's fall-out to a suspect process level; the inability to distinguish pre-packaging yield issue from packaging yield issues; and the inability to establish a clear link between large populations of failed die.
Failure analysis may make use of a known electrical diagnosis process whereby a diagnostic list of suspected failing nets may be obtained as shown in FIG.
2
. Packaged devices having BIST (scan) capabilities are tested using a tester. Scan failure data is translated into format that can be used by an ATPG (Automatic Test Pattern Generation) tool, e.g., an ATPG tool used previously to generate test pattern files used by the tester. The ATPG tools use the translated test data, together with the test pattern files, setup files, one or more ATPG diagnostic models, and design information from a design database in order to identify suspected failing nodes, output in the form of a diagnostic list (FIG.
3
).
CAD navigation tools have been developed to aid in failure analysis. CAD navigation refers to the ability to point and click within a circuit layout display and by so doing automatically drive a piece of equipment such as FIB (Focussed Ion Beam) equipment to that location on the chip. CAD navigation also allows a user to specify the name of a net, causing the corresponding layout to be displayed. Once such CAD navigation tool is the Merlin Framework CAD navigation tool of Knights Technology. This tool takes netlist information, layout information, and cross-reference files relating the two and procedures a unified database in a Knights-proprietary format having an efficient indexing structure. Referring more particularly to
FIG. 4
, the process of creating such a Knights database is illustrated in greater detail. A SPICE-formatted netlist is converted (if required) to a suitable netlist format. Data from a Schematic Verification database is converted (if required) to obtain cross-reference files cross-referencing net names and numerical net identifiers. Layout data is converted (if required) to a suitable polygon layout format. An MMapper routine uses the netlist, cross-reference files and polygon layout files in the following manner to produce a database suitable for CAD navigation.
In a co-pending application Ser. No. 09/192,164 filed Nov. 13, 1998 (now U.S. Pat. No. 6,185,707B1), of Shawn Smith, Han Balachandran and Jason Parker entitled, “IC Test Software System for Mapping Logical Functional Test Data of Logic Integrated Circuits to Physical Representation,” incorporated herein by reference, takes advantage of the foregoing capability to determine and display the X,Y location corresponding to a net name, by translating functional test data of a digital logic chip passed through a simulation model which identifies one or more defective nets of the chip. The defective nets are processed against a database of the type to obtain X, Y coordinate data for these nets, allowing them to be data tagged as physical traces on the chip layout. The mapping is performed by taking the output from a functional tester and translating it form a list of failed scan chains into a list of suspected netlist nodes. The X, Y coordinates of suspected netlist nodes are then identified and stored in a database, providing
Nghiem Michael
Telecky , Jr. Frederick J.
Texas Instruments Incorporated
Troike Robert L.
LandOfFree
System for mapping logical functional test data of logical... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System for mapping logical functional test data of logical..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System for mapping logical functional test data of logical... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3075098